EVT in Electricity Price Modeling: Extreme Value Theory not only on the Extreme Events

Z. Marossy (Hungary)


Risk analysis, extreme value theory, electricity prices


My model is an unconditional model according to [1]. This means that the data are not filtered before the analysis. Extreme value theory (EVT) is used to model extreme electricity prices in the literature. In this paper I show that EVT can be used to describe the distribution of the prices itself. I build a model on the price formation in electricity auctions markets. The conclusion is that the distribution of electricity prices should be a generalized extreme value (GEV) distribution. Empirical data confirm this conclusion. In the literature there are three main modeling approaches to describe the distribution of electricity prices. The models in the first model family fit a stochastic model on the electricity price time series and use this stochastic model to generate the given distribution (see [5]). The models of the second group consist fundamental descriptions of electricity demand and supply, and the driving factors of supply and demand determine the price distribution (see [6]). The third type of models is agent-based models which use simulation techniques to write down the price distribution (see [7]). This paper contains a fourth way of research. It aims to determine the distribution of electricity prices directly without knowing anything about the data generating process or market driving forces.

Important Links:

Go Back